English

应用科学学报 ›› 2003, Vol. 21 ›› Issue (4): 353-356.

• 论文 • 上一篇    下一篇

基于MAP算法的无教师讲者自适应的研究

赵力, 邹采荣, 吴镇扬   

  1. 东南大学无线电工程系 江苏 南京 210096
  • 收稿日期:2002-09-11 修回日期:2003-03-25 出版日期:2003-12-10 发布日期:2003-12-10
  • 作者简介:赵力(1958-),男,江苏南京人,教授,博士;邹采荣(1961-),男,江苏昆山人,教授,博导;吴镇扬(1948-),男,江苏扬州人,教授,博导.
  • 基金资助:
    国家自然科学基金资助项目(60272044)

A Study on the Chinese Continuous Speech Recognition Unsupervised Speaker Adaptation Based on Maximum a Posteriori Algorithm

ZHAO Li, ZOU Cai-rong, WU Zhen-yang   

  1. Department of Radio Engineering, Southeast University, Nanjing 210096, China
  • Received:2002-09-11 Revised:2003-03-25 Online:2003-12-10 Published:2003-12-10

PDF

89

被引次数

3

摘要： 提出了一种基于MAP算法的连续语音识别无教师讲者自适应方法,给出了简单有效的渐进自适应公式.通过识别实验结果表明,提出的方法能利用少量的自适应数据达到满意的学习效果,是一种实用和有效的讲者自适应方法.

关键词: 无教师讲者自适应, 汉语连续语音识别, MAP算法

Abstract: Aiming at Chinese continuous speech recognition, a novel unsupervised speaker adaptation method based on maximum a posteriori algorithm is presented in this paper, together with a simple effective formula which can give a satisfying result with only a few data. It has been proved to be an effective method to apply.

Key words: recognition of Chinese continuous speech, unsupervised speaker adaptation, maximum a posteriori algorithm

中图分类号: